MQA vs HiRez: an apples-to-apples comparison

[fas42]

Idea of what?

[kumakuma]

5 minutes ago, fas42 said:

Idea of what?

 

you're talking about

 

 

[fas42]

Hmmm ... the one clear takeway from this MQA mess is that hardly anyone knows what's going on - so I'm in good company there. I'm attempting to get some idea of what this bit of nonsense is doing, fiddling with the sound, and use those snippets which are readily available which purport to demonstrate the difference in the sound - I have zero interest in paying money for the privilege.

 

As

@mansr

 points out, MQA is Yet Another DSP fiddle - up to the user whether they want to pay constantly for use of such.

 

BTW, the CA editor is a nutter job - just trying to get the mansr tag in the last paragraph to take in the last paragraph was a fight - why can't people get software to work better ...

[PeterSt]

59 minutes ago, kumakuma said:

I'm confused. You said you didn't have a DAC and only listen to CDs. How are you listening to MQA releases?

 

47 minutes ago, kumakuma said:

You really have no idea at all.

 

Maybe this is not fair.

It is easy to fall in the pitfall of the impossibility to use a laptop, not Roon and not XXHigHend and thinking : Okay, he thus did not listen to MQA. But he did because indeed the "unfolded" versions can be put up and thus is MQA to its fullest extent, you adding your own further upsampling filtering, or not.

 

It would be true that nobody can have an idea at all in this short notice.

Did I decide yet ?

[fas42]

Peter, it's trivially easy to hear the differences, as many have stated, even on a "pathetic" laptop - that is, assuming that we are listening to what MQA is doing. The waveform is different, visually obviously so - case closed ...

[kumakuma]

22 minutes ago, fas42 said:

BTW, the CA editor is a nutter job - just trying to get the mansr tag in the last paragraph to take in the last paragraph was a fight - why can't people get software to work better ...

 

Just type "@" and then the first letters of his handle and a list of candidates appears. The more letters you type, the shorter the list becomes.

[fas42]

8 minutes ago, kumakuma said:

 

Just type "@" and then the first letters of his handle and a list of candidates appears. The more letters you type, the shorter the list becomes.

 

Yes, that's how it normally, and should work - but this time I could not get the cursor to leave the selected tag. It refused to move to where it should, type Enter, Space, what's the magic keystroke it's decided it needs ...? Trouble is, programmers assume what people do, and if you deviate the tiniest fraction from that scenario, all bets are off ... you have to train the user, .

[manisandher]

10 hours ago, crenca said:

Yes, but all PCM on Tidal is 16/44....  

 

Yes, but that's not what you said:

 

On 10/18/2018 at 4:28 PM, crenca said:

Bob James' "The New Cool" is one (in 48) that is available in MQA and 48 PCM on Tidal...

 

In any event, I'll look around for some material, check that the MQA and PCM are both from the same master and then take some captures.

 

Mani.

[crenca]

8 hours ago, manisandher said:

 

Yes, but that's not what you said:

 

 

In any event, I'll look around for some material, check that the MQA and PCM are both from the same master and then take some captures.

 

Mani.

 

Oops, sorry I was not clear!  I must have been thinking "MQA 48 PCM equivalent master"... 

[vortecjr]

I know a little bit about Roon and Audacity.

 

First, there are two settings you need to worry about in Roon:

MQA capabilities (drop down)
MQA core decoding (slider)

 

According to Roon...

Drop down is where the user tells Roon what their dac can do. Based on the capabilities, Roon makes the "best" decision for every situation, i.e. lets the DAC do as much of the process as it can support given the user's settings.

 

The slider is a kill switch for anti-mqa people who just want to be super sure that Roon plays MQA bit-perfect and undecoded no matter what.

 

Assuming Roon is sending PCM to Audacity...Audacity is capturing the stream based on capture settings which are a user input. These capture settings should match the sample rate, bit rate, channels, etc. of the source stream. This creates a wave form you can analyze in the program. Remember Audacity does not have any clue about MQA. At this point you have captured the data and created a wave form in what is called a Project. Next you can Export the project to one of a few audio formats. This export function also has user settings for sample rate, bit rate, channels, dither, etc. During this process you can also add metadata because none will be transferred through the capture process. 

 

I have to agree with Mansr. There is a lot of room for unexpected / unintentional things to change along the way and alter the end result. 

[Miska]

Just posted on the other thread, if someone wants to compare:

 

So far, my favorite comparison album has been:

https://www.highresaudio.com/en/album/view/643ush/fresh-moods-peter-haubfleisch-orfine

 

Especially the track 7 there, since it has some properties that push the MQA encoder a bit more.

[mansr]

3 minutes ago, Miska said:

So far, my favorite comparison album has been:

https://www.highresaudio.com/en/album/view/643ush/fresh-moods-peter-haubfleisch-orfine

 

Especially the track 7 there, since it has some properties that push the MQA encoder a bit more.



So much that it uses 9 bits per sample, rather than the usual 8, to encode the high half of the spectrum.

[Miska]

5 minutes ago, mansr said:

So much that it uses 9 bits per sample, rather than the usual 8, to encode the high half of the spectrum.

 

Yes, and while doing so needs to cut the baseband resolution even more... I spent quite a bit of of time trying to find MQA album that would actually have notable amount of ultrasonic content to push the bit allocator. Because I expected that such material would expose some more differences, and maybe some peak compression and noise modulation.

 

Because MQA's presentations have only included expected spectrum slope of classical music. And most demo material I've seen is classical or something else that is not so "busy". This one has less steep downward slope.

 

Would be interesting to find some hires heavy metal that would have massive "wall of sound" throughout the spectrum. And maybe some drummer tracks with lot of different percussions with multi-track close-mic recording.

 

Testing these things is tricky because one cannot convert their own content to MQA for comparison... That would be needed for basic acceptance testing.

 

[mansr]

6 minutes ago, Miska said:

Testing these things is tricky because one cannot convert their own content to MQA for comparison... That would be needed for basic acceptance testing.

Which is why they won't let you do that.

[NoisyNarrowBandDevice]

7 hours ago, Miska said:

Would be interesting to find some hires heavy metal that would have massive "wall of sound" throughout the spectrum. And maybe some drummer tracks with lot of different percussions with multi-track close-mic recording.

 

 

Krisiun's latest album "Scourge of the Enthroned" might fit the bill. There is a 24/48 version available but can somebody please check if its on Tidal in MQA? The Tidal web-player search function gives no indication.

 

 

 

[manisandher]

8 hours ago, vortecjr said:

I know a little bit about Roon and Audacity.

 

First, there are two settings you need to worry about in Roon:

MQA capabilities (drop down)
MQA core decoding (slider)

 

According to Roon...

Drop down is where the user tells Roon what their dac can do. Based on the capabilities, Roon makes the "best" decision for every situation, i.e. lets the DAC do as much of the process as it can support given the user's settings.

 

The slider is a kill switch for anti-mqa people who just want to be super sure that Roon plays MQA bit-perfect and undecoded no matter what.

 

Assuming Roon is sending PCM to Audacity...Audacity is capturing the stream based on capture settings which are a user input. These capture settings should match the sample rate, bit rate, channels, etc. of the source stream. This creates a wave form you can analyze in the program. Remember Audacity does not have any clue about MQA. At this point you have captured the data and created a wave form in what is called a Project. Next you can Export the project to one of a few audio formats. This export function also has user settings for sample rate, bit rate, channels, dither, etc. During this process you can also add metadata because none will be transferred through the capture process. 

 

I have to agree with Mansr. There is a lot of room for unexpected / unintentional things to change along the way and alter the end result. 

 

Not sure if you've read the whole thread, but here's how Roon is configured with the MQA file:

 

On 10/16/2018 at 4:35 PM, manisandher said:

1. Roon signal path - [DAC has] no MQA support

 

And I have Audacity set up correctly:

 

On 10/17/2018 at 4:17 PM, pkane2001 said:

An under -190dB difference and a 200dB correlated null depth are as perfect a match as one can do with the precision of calculations used in DeltaWave.

 

Mani.

[PeterSt]

13 hours ago, Miska said:

Would be interesting to find some hires heavy metal that would have massive "wall of sound" throughout the spectrum

 

Miska, last year I tried my best to find Heavy Metal MQA material. This is/was not all that easy. In the view below you see three which made it for at least one track to my "Demo" folder.

 

 

That would be the Iron Maiden, the Megadeth and the Motörhead.

As you can see, only the Iron Maiden is "Hires".

 

The DR numbers are nothing short from shocking :

 

If that's not implying a wall of sound then I don't know what is.

[Miska]

3 minutes ago, PeterSt said:

Miska, last year I tried my best to find Heavy Metal MQA material. This is/was not all that easy. In the view below you see three which made it for at least one track to my "Demo" folder.

 

 

That would be the Iron Maiden, the Megadeth and the Motörhead.

As you can see, only the Iron Maiden is "Hires".

 

The DR numbers are nothing short from shocking :

 

If that's not implying a wall of sound then I don't know what is.

 

Thanks! That Iron Maden is listed on HDtracks only and not available to Finland. On highresaudio.com there for example this:

https://www.highresaudio.com/en/album/view/xhare6/anthrax-kings-among-scotland-live

https://www.highresaudio.com/en/album/view/ou9mx6/northward-northward

But it is not available to Finland either. Overall, for example nothing published by Nuclear Blast is available to Finland.

 

This sounds sonically promising, but it is not hires:

https://www.highresaudio.com/en/album/view/vgjrbx/disturbed-evolution-deluxe

 

Would be curious to see how peak spectrum of the Iron Maden looks like. If they have good gear it should extend high level throughout the ultrasonic range. Old stuff that has gone through analog tape won't work.

[PeterSt]

22 minutes ago, Miska said:

Would be curious to see how peak spectrum of the Iron Maden looks like.

 

This is what I can do quickly from Track 09 - The Mercenary (the track in my Demo folder) :

 

   

 

Looks to extend to close to 25KHz. But watch the line close to 20KHz as well in a quiet (intro) track where you can see this :

 

    

 

[PeterSt]

@Miska, I copied the wrong file at first ... Please look again in my previous post.

[Miska]

7 minutes ago, PeterSt said:

 

This is what I can do quickly from Track 09 - The Mercenary (the track in my Demo folder) :

 

   

 

Looks to extend to close to 24KHz. But watch the line close to 20KHz as well.

 

 

Yeah, either analog mix from 48 kHz ADC/DAC or some analog tape on the way...

[pkane2001]

8 hours ago, manisandher said:

 

Not sure if you've read the whole thread, but here's how Roon is configured with the MQA file:

 

 

And I have Audacity set up correctly:

 

 

Mani.

 

Hi Mani, just want to make sure you are quoting the right stats from my posts. The 200dB null was between the two non-MQA captures. The difference between MQA and non-MQA was around 60dB. Still  small, but in the audible range.

[manisandher]

15 minutes ago, pkane2001 said:

Hi Mani, just want to make sure you are quoting the right stats from my posts. The 200dB null was between the two non-MQA captures. The difference between MQA and non-MQA was around 60dB. Still  small, but in the audible range.

 

Yes. When someone queries whether I have Audacity set up correctly, the only stat of interest is the null between the non-MQA original and the non-MQA capture, which you measured as -200dB or so. This proves that Audacity is set up optimally.

 

As I've said before, I would be surprised if the null between the non-MQA original and the MQA capture were anywhere near -200dB. A null of around -60dB suggests to me that there is some additional processing in the MQA encode/decode over and above a simple first unfold. Exactly what this processing is is what I'm interested in, and hopefully the new apples-to-apples thread will prove insightful here.

 

Mani.

[mansr]

9 minutes ago, manisandher said:

Yes. When someone queries whether I have Audacity set up correctly, the only stat of interest is the null between the non-MQA original and the non-MQA capture, which you measured as -200dB or so. This proves that Audacity is set up optimally.

No. The fact that there is any difference at all proves that something is altering the data. We have determined that it is most likely Audacity applying dither. You really ought to disable that for this test.

[manisandher]

11 minutes ago, mansr said:

No. The fact that there is any difference at all proves that something is altering the data. We have determined that it is most likely Audacity applying dither. You really ought to disable that for this test.

 

I said right from the beginning that dither was set to 'off' in Audacity. If it is still applying dither irrespective, would you expect a null of around -200dB between the original and the non-MQA capture?

 

Mani.